|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"global_step": 795, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 6.25e-07, |
|
"loss": 0.856, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.25e-06, |
|
"loss": 0.792, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.8750000000000003e-06, |
|
"loss": 0.835, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 2.5e-06, |
|
"loss": 0.8315, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 3.125e-06, |
|
"loss": 0.8115, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 3.7500000000000005e-06, |
|
"loss": 0.8032, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.3750000000000005e-06, |
|
"loss": 0.7256, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 5e-06, |
|
"loss": 0.7007, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 5.625e-06, |
|
"loss": 0.647, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.25e-06, |
|
"loss": 0.6431, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.875e-06, |
|
"loss": 0.6455, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 7.500000000000001e-06, |
|
"loss": 0.5786, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 8.125000000000001e-06, |
|
"loss": 0.6245, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 8.750000000000001e-06, |
|
"loss": 0.5962, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.375000000000001e-06, |
|
"loss": 0.5854, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1e-05, |
|
"loss": 0.5693, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.0625e-05, |
|
"loss": 0.5615, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.125e-05, |
|
"loss": 0.5354, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.1875e-05, |
|
"loss": 0.563, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.25e-05, |
|
"loss": 0.5596, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.3125e-05, |
|
"loss": 0.543, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.375e-05, |
|
"loss": 0.5403, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.4375e-05, |
|
"loss": 0.5437, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.5000000000000002e-05, |
|
"loss": 0.5286, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.5625e-05, |
|
"loss": 0.5417, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.6250000000000002e-05, |
|
"loss": 0.5151, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.6875e-05, |
|
"loss": 0.5154, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.7500000000000002e-05, |
|
"loss": 0.5271, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.8125e-05, |
|
"loss": 0.5283, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.8750000000000002e-05, |
|
"loss": 0.5127, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9375e-05, |
|
"loss": 0.5125, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 2e-05, |
|
"loss": 0.5166, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9999915234318064e-05, |
|
"loss": 0.5288, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.999966093870929e-05, |
|
"loss": 0.5264, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.99992371174848e-05, |
|
"loss": 0.5229, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9998643777829674e-05, |
|
"loss": 0.5239, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9997880929802895e-05, |
|
"loss": 0.4995, |
|
"step": 37 |
|
}, |
|
{ |
|